Endogenous galectin-1 exerts tonic inhibition on experimental arthritis

Abstract

Little is known about the role(s) of endogenous Gal-1 during arthritis. In this study we queried whether anti-arthritic functions for this effector of endogenous anti-inflammation could be unveiled, by studying collagen-induced arthritis (CIA) in Gal-1^−/− mice.

Gal-1^−/− and C57BL/6J (WT) mice received an immunisation of chicken type II collagen (CII) in complete Freund’s adjuvant (FA) followed by a booster on day 21, which consisted of CII in incomplete FA. Animals were monitored for signs of arthritis from day 14 onwards. Clinical and histological signs of arthritis were recorded and humoral and cellular immune responses against CII were analysed.

A distinct disease penetrance was apparent, with ~70% of Gal-1^−/− mice developing arthritis compared to ~50% in WT animals. Gal-1^−/− mice also exhibited an accelerated disease onset and more severe arthritis characterised by significantly elevated clinical scores. Post-mortem analyses (day 42) revealed higher levels of IgG1 and IgG2b anti-CII immunoglobulin isotypes in the serum of Gal-1 null animals compared to WT. Lastly, T cell responses following ex-vivo stimulation with CII revealed a greater degree of proliferation in T cells of Gal-1^−/− mice compared to WT, which was associated with increased production of IL-17 and IL-22.

These data provide the novel notion that endogenous Gal-1 is an inhibitory factor in the development of arthritis affecting disease severity. We have also highlighted the importance of endogenous Gal-1 in regulating T cell reactivity during experimental arthritis.

INTRODUCTION

Despite successes achieved by the introduction of biologics for the management of Rheumatoid Arthritis (RA), up to a third of patients fail to respond adequately to anti-TNF therapy (1, 2). There is a need for novel pharmacological or immunoregulatory approaches, with novel molecules to be used, perhaps, in association with current therapies. Galectins may provide a novel opportunity in the development of anti-RA therapeutics with profound immuno-regulatory effects for members of this family of glycan binding proteins identified (3).

Among the 15 Galectins, our own work has focused on Galectin-1 (Gal-1), starting from the observation that treatment of mice with small doses of this protein elicits potent inhibitory effects on the process of leukocyte recruitment in inflammation (4, 5). These data have been extended from mouse to human systems indicating the potential translational relevance of this line of research (6, 7). More recently we have complemented these pharmacological analyses by addressing the patho-physiological role of endogenous Gal-1 using the Gal-1^−/− mouse (8) in a model of paw oedema (9). Against expectations, oedema was significantly diminished in the absence of endogenous Gal-1 (9), a finding that highlights the importance of studying the role of endogenous galectins in experimental models of disease.

Limited information exists regarding Gal-1 and arthritis. In a seminal study (10), delivery of Gal-1 via secretion by syngeneic fibroblasts reduced the clinical severity of CIA. Such marked macroscopic effect was associated with induction of apoptosis in Th1 cells. This conclusion was corroborated by the observation that local intra-articular delivery of a lentiviral vector over-expressing Gal-1 ameliorated clinical CIA in rats (11). Whilst these two studies indicate potential for the pharmacological administration of Gal-1 they shed little light on the potential properties of the endogenous protein on the processes characteristic of experimental CIA. In the present study we addressed this issue by monitoring Gal-1 expression in arthritic joints of DBA/1 mice and subsequently tested the Gal-1^−/− mouse in a protocol adapted to the C57BL/6J strain. We show that absence of endogenous Gal-1 results in an exacerbation of disease, both in terms of incidence and clinical severity.

MATERIALS AND METHODS

Animals

Male DBA/1 mice (Harlan, Oxfordshire, UK) and female C57BL/6J (WT) and Gal-1^−/− mice (Charles River, Kent, UK) were used at 12-15 weeks old under Home Office UK regulations. Original breeding pairs of Gal-1^−/− (Lgals1 null) animals on a C57BL/6J background were provided by the Consortium for Functional Glycomics (http://www.functionalglycomics.org). These mice were verifed as being 88% C57BL/6J by a 384 SNP panel analysis performed by Charles River, USA.

Collagen-induced arthritis

Initial experiments were performed to assess galectin expression in the classical model of CIA performed in the DBA/1 strain as previously described (12). Briefly, mice were immunised with 100 μg chicken type II collagen (CII; mdbioproducts, Zurich, Switzerland) in Complete Freund’s adjuvant (CFA) followed by a booster injection using incomplete Freund’s adjuvant (IFA) on day 14. Mice were examined daily for signs of arthritis.

CIA was induced in Gal-1^−/− and WT mice (C57BL/6J background) as recently described (13). Briefly, mice were injected with 200 μg CII in CFA followed by a booster injection on day 21 in IFA. Mice were scored daily from day 14 for clinical signs of arthritis (0-3 scale taking into account swelling of the wrist/ankle, pad and digits); edema was assessed by plethysmometry. Serum and draining lymph nodes were collected from mice at early (day 26, coinciding with disease onset in the majority of animals) and late (day 41, established arthritis in both groups) time-points for further analyses.

Real-time PCR

RNA was extracted from homogenised joints using RNeasy Plus mini-kit (Qiagen, Sussex, UK). Real-time PCR was performed using primers from a commercial source (Qiagen). Ct values were normalised to endogenous Gapdh and data were analysed using the 2^−(ΔΔCt) method.

Cell culture

Published protocols (13) were used to analyse T cell responses in mice subjected to CIA. Draining inguinal lymph node (LN) cells were collected at day 26 and 41 and stimulated with CII (50 μg/ml) or anti-CD3 (eBioscience, Wembley, UK) (100 ng/ml) for 48 h. IL-17, IL-22 and IFN-γ levels were measured by ELISA following manufacturer’s guidelines (R&D systems, Oxford, UK) and cell proliferation was assessed by ³H thymidine incorporation as previously described (14).

ELISA

Serum IgG levels were detected with anti-mouse IgG1, and/or IgG2b HRP- conjugated Abs diluted 1:5000 (AbD Serotec, Abingdon, UK). Serum amyloid A (SAA) levels were measured by ELISA along manufacturer’s guidelines (Tridelta Development Ltd).

Histology

5 μm sections of paws and joints were stained with H&E or safranin O/fast green (Sigma-Aldrich). Specific immunostaining was performed for Gal-1 using polyclonal goat-anti-mouse Gal-1 (R&D systems). A minimum of 4 sections per animal were evaluated.

Flow cytometry

Freshly harvested draining LN cells were stained with FITC-conjugated rat-anti mouse CD4, PE-conjugated rat-anti mouse CD8, PE-Cy5 conjugated rat-anti mouse CD44, APC-conjugated rat-anti mouse CD62L and isotype controls (eBR2a). Cell populations were analysed with a FACScalibur flow cytometer using CellQuest software (BD Bioscience, Oxford, UK).

Statistical Analysis

Data are expressed as mean±SEM and analysed using one-way analysis of variance, t-test or Area Under the Curve followed by the Student’s t-test. P value ≤ 0.05 were considered significant.

RESULTS

Characterisation of endogenous galectin expression in arthritic joints

In DBA/1 mice, clinical signs of arthritis were first observed by day 28 (mean clinical score 0.86±0.26) which correlated with peak expression of Saa in ankle and wrist joints (Figure 1A). By day 40, 80% of mice displayed signs of arthritis (mean clinical score 3.47±0.65).

Figure 1. Selected mediator expression in arthritic joints of DBA/1 mice.

Male DBA/1 mice were induced with CIA and quantitiative real-time PCR was performed on cDNA from ankle and wrist joints collected across the course of disease. (A) Saa, (B) Tnf, (C) Il6, (D) Il17, (E) Il10, (F) Tgfβι. Data are expressed as mean ± SEM (n= 6) *P<0.05 vs. control (control joints were taken from non-immunized mice).

The mRNA for the pro-inflammatory cytokines Tnf, Il6 and Il17a was observed as early as day 6, peaking at day 28 (Figure 1B-D). Anti-inflammatory cytokines Il10 and Tgfb were also monitored: Il-10 mRNA levels were downregulated (up to day 15) and then returned to basal values (day 28; Figure 1E) whilst Tgfb mRNA expression peaked at day 28 (Figure 1F). Analysis of the same cytokines in C57BL/6J and Gal-1^−/− mice revealed upregulation of Saa, Il6, Il10, Tgfb and Il4 at day 42 with only Saa levels significantly different between genotypes (supplementary Figure 1).

mRNA for Gal-1 and Gal-9 was strongly modulated with peak expression correlating with clinical onset of disease (day 28) and remaining elevated up to day 40 (peak clinical disease) (Figure 2A). Gal-3 mRNA expression was not modulated throughout this time-course (Figure 2A).

Figure 2. Gal-1 expression in arthritic joints of DBA/1 mice.

Male DBA/1 mice were induced with CIA and quantitiative real-time PCR was performed on cDNA from ankle and wrist joints collected across the course of disease. (A) Lgals1 (Gal-1), Lgals3 (Gal-3) and Lgals9 (Gal-9) expression. Data are expressed as mean ± SEM (n= 6) (B, C, D, E) Gal-1 expression in cartilage and bone at day 28, (F, G) Gal-1⁺ cells/synoviocytes and at sites of invasion. (H, I) Gal-1⁺ infiltrated leukocytes. E, G and I show higher magnification of boxed areas in D, F and H respectively. (Scale bars B, D, F, H: 200 μm; C, E: 100 μm; G, I: 80 μm). Ct, connective tissue; PB, periosteum; EB, endochondral bone; MC, marrow cavity; Oc, osteoclasts; JC, joint cavity/space; C, cartilage.

Given the strong degree of modulation of Gal-1 and Gal-9 at mRNA level, expression was monitored by immunohistochemistry in joints with maximal clinical score. Gal-1 expression was intense in the synovial lining and throughout the cellular infiltrate (Figure 2B-G). Concentrated staining was also observed in osteoclasts (Figure 2D) and chondrocytes located within the proliferative zone of the cartilage (Figure 2F & G). Gal-1 positive infiltrating leukocytes could also be observed (Figure 2H & I). Gal-9 staining appeared to be restricted to fibroblast like synoviocytes and osteoblasts (data not shown). Collectively, these initial analyses prompted us to test the potential functional relevance of endogenous Gal-1.

Absence of endogenous Gal-1 leads to increased disease incidence and severity

A significantly larger proportion Gal-1^−/− mice (69%) developed arthritis compared to WT (53%) (Figure 3A & B). The mean day of onset was also earlier in Gal-1^−/− mice. In terms of clinical manifestations, Gal-1^−/− mice displayed enhanced disease severity (Figure 3C & D), with a mean clinical score of 4.57±1.28 compared to 2.89±0.93 in WT mice. Edema was mildly elevated in the Gal-1^−/− mice compared to WT mice (Figure 3E & F). Histological analysis of joints taken at days 31 and 42 from both WT and Gal-1^−/− mice revealed a modest elevation in score in Gal-1^−/− mice that was not statistically significant (data not shown).

Figure 3. Increased incidence and severity of CIA in Gal-1^−/− mice.

Female WT (C57BL6/J) and Gal-1^−/− mice were induced with CIA and monitored daily for signs of arthritis. (A) incidence, (C) clinical score (0-3 scale taking into account swelling of the wrist/ankle, pad and digits; maximum score 12 per mouse) and (E) level of oedema. AUC analyses for (B) incidence, (D) clinical score and (F) edema. Data are expressed as mean ± SEM (n= 15 WT; n=13 Gal-1^−/−). *P<0.05, **P<0.01 vs. WT.

Elevated SAA and anti-collagen IgG levels in Gal-1^−/− mice

Elevated levels of SAA were observed in both groups by day 26 (disease onset, Figure 4A). Interestingly, on day 42, SAA levels in Gal-1^−/− mice increased further and were significantly higher than those observed in WT animals. Coupled with the clinical data, this result suggests absence of endogenous Gal-1 may promote joint inflammation during the progression/active phase of the disease.

Figure 4. Circulating levels of serum Amyloid A (SAA) and anti-CII Abs in WT and Gal-1^−/− mice during CIA.

Serum was collected at disease onset (day 26) and during established disease (day 42) as well as from from naïve (n=2) WT mice at day 42. Levels of (A) SAA, (B) IgG1 and (C) IgG2b were detected by ELISA. Data are expressed as mean ± SEM. (day 26 n= 3, day 42 n= 10 WT & 7 Gal-1^−/−). **P<0.01, ***P<0.001 vs. WT (day 42).

Collagen specific IgG levels serve as markers of Th1/Th17 and Th2 responses respectively with IgG1 indicative of a Th2 response, IgG2a and b of a Th1 response, whilst Th17 cells have been shown to increase class switch recombination to IgG1, IgG2a, IgG2b and IgG3 (15). IgG2a levels were not measured as the gene encoding this isotype is deleted in C57BL/6 mice (16). IgG1 levels in Gal-1^−/− mice increased across the time-course and were significantly elevated at day 42 (Figure 4B). IgG2b increased in both genotypes over the course of the disease with significantly higher levels detectable at day 42 in the Gal-1^−/− mice (Figure 4C).

Arthritic Gal-1^−/− mice show increased T cell effector responses, proliferation and skewing towards a Th17 phenotype

In response to stimulation with CII, Gal-1^−/− LN cells exhibited a significantly higher proliferative response compared to WT cells (Figure 5A and B). Furthermore, CII stimulation led to significantly higher levels of IL-17 and IL-22 being produced by Gal-1^−/− cells compared to WT at both onset and established stages of disease, day 26 and 42, respectively (Figure 5C and D). Comparable levels of IFN-γ were recovered from both WT and Gal-1^−/− cells harvested at day 26 (data not shown).

Figure 5. Enhanced proliferation and generation of IL-17 and IL-22 by Gal-1^−/− T cells in response to specific antigen.

Draining LN cells were isolated from animals at disease onset (day 26) and during established disease (day 42) and stimulated with CII or CD3 for 48h. LN were also collected from naïve (n=2) WT mice. Proliferation was assessed by incorporation of tritiated thymidine (A) day 26 and (B) day 42. Data are expressed as mean ± SEM, (day 26 n=3, day 42 n= 10 WT and 7 Gal-1^−/−). *P<0.05 vs. WT. Levels of (C) IL-17 and (D) IL-22 in cell supernatants were measured by ELISA. Data are expressed as mean ± SEM (n= 5-10 per time-point). *P<0.05 vs. WT.

At day 42 the percentage of CD4⁺ and CD8⁺ T cells in draining LN was proportional in WT mice in contrast to Gal-1^−/− mice where double the percentage of CD4⁺ T cells was found (Figure 6A & B). In terms of T cell counts, Gal-1^−/− mice displayed double values of CD4⁺ T cells compared to WT (Figure 6E). Similar numbers of CD8⁺ T cells were observed in both strains (data not shown). Additional staining was carried out on the CD4⁺ population for CD44 and CD62L (Figure 6C & D) in order to identify the percentage of effector memory T cells. Figure 6F shows that the percentage of effector memory CD4⁺ T cells (CD44⁺CD62L^low) was significantly increased in the Gal-1^−/− mice compared to WT.

Figure 6. T cell priming and expansion in the absence of endogenous Gal-1.

Representative dot plots displaying the percentage of CD4/CD8 levels in draining LN of (A) WT and (B) Gal-1^−/− mice at day 42. Representative dot plots displaying the proportion of CD62L/CD44 levels on CD4⁺ T cells for (C) WT and (D) Gal-1^−/− mice. (E) Total numbers of CD4⁺ T cells and (F) percentage of CD4⁺CD44⁺CD62L^low effector memory cells in the two gentoypes at day 42. Data are expressed as mean ± SEM (n= 7 WT; n=5 Gal-1^−/−). *P<0.05, **P<0.01 vs. WT.

Discussion

The effects of galectins are complex and vary depending on their cellular localisation, it is therefore important to consider the actions of both the endogenous and exogenous protein in order to fully appreciate their biology. This study demonstrates, for the first time in a model of CIA, that absence of endogenous Gal-1 leads to enhanced susceptibility and disease severity, which is in part associated with exacerbated T cell expansion and skewing towards a Th17 profile.

The first set of results reported mRNA expression of several cytokines in joints taken from DBA/1 mice during the course of CIA. As expected, levels of the pro-inflammatory cytokines IL-6, TNFα and IL-17 were elevated, all of which have been implicated in the pathogenesis of both RA and CIA (17-20). Interestingly endogenous IL-10 mRNA levels remained below control threshold, indicating a down-modulation in gene expression. We propose that during onset and active phases of CIA the immunosuppressive actions of IL-10 are overun by the plethora of pro-inflammatory cytokines present in the joint.

In RA, galectins display a broad pattern of expression within the synovium (21, 22). However, to the best of our knowledge this is the first study to investigate Gal-1 expression over the course of CIA in mice. In line with other inflammatory models such as EAE (23), Gal-1 mRNA expression peaked at the clinical onset of disease and remained elevated during disease progression which may be indicative of a pro-resolutory role for the protein and may correlate with an increased expression in activated T cells recruited to the joint. Immunohistochemistry of arthritic joints indicated strong co-localisation between Gal-1 and infiltrating synovial cells with a strong degree of immunoreactivity also in chondrocytes, a result in-line with the recent description of Gal-1 as a novel factor in chondrocyte growth, cartilage formation and maturation (24). Whilst numerous Gal-1 positive leukocytes were observed in the synovium it is not clear whether these cells express Gal-1 or have taken it up from the extracellular environment. Peripheral blood cells express little or no Gal-1 whilst activated T cells upregulate Gal-1 over a period of days (25). Recent evidence has shown that another galectin (Gal-3) is endocytosed by macrophage-like cells (26), whilst neutrophils recovered from BAL fluid of strep pneumoniae infected mice have increased levels of Gal-3 bound extracellularly (27).

The action of Gal-1 in models of inflammation signals a dominant anti-inflammatory role for this protein (3, 5). Recent data suggests several mechanisms through which Gal-1 may be protective in chronic autoimmune pathologies including RA and CIA. These would include induction of apoptosis of Th1 and Th17 effector cells (23), antagonism of T cell activation and proliferation (28-30), enhanced immune tolerance through positive effects on Treg development and function (31, 32) and induction of IL-10 synthesis in activated Th cells (33, 34).

The significant increase in proliferation observed in Gal-1^−/− cells in response to the specific antigen CII supports an anti-proliferative role for Gal-1 in response to antigen stimulation (28). Further characterisation of CD4+ T cells recovered at day 42 revealed a greater proportion of CD4⁺ effector memory T cells in the lymph nodes of Gal-1^−/− mice compared to WT. Clearly absence of endogenous Gal-1 appears to have a significant impact on the regulation of T cell responses, which could, in part, explain the disease profile observed in Gal-1^−/− mice.

IL-4 secreting Th2 cells have been demonstrated to support IgG1 production by plasma cells and IL-17 in combination with IL-21 generated by Th17 cells supports class switching and production of IgG1 and IgG2b isotypes (15, 35, 36). Levels of both IgG1 and IgG2b were elevated in Gal-1^−/− mice suggesting the establishment of a dominant Th17 immune response. This was further supported by increased production of IL-17 and IL-22 by draining LN cells of Gal-1^−/− mice in response to CII. Th17 cells induce tissue inflammation and drive disease pathogenesis in a variety of experimental autoimmune diseases including EAE (37) and CIA (19, 38) as well as RA (39) and multiple sclerosis (40) in humans. A role for IL-22 in driving CIA in C57/BL6 mice has also been demonstrated (41). Strong evidence in the literature demonstrates a role for Gal-1 in dampening Th1 and Th17 mediated responses and skewing toward a Th2 profile. An elegant study by Toscano et al (23) showed Th1 and Th17 cells express a specific repertoire of cell-surface glycans, which increases their susceptibility to Gal-1-induced cell death; in contrast Th2 cells are protected due to differential α2-6 sialylation of cell surface glycoproteins. This could explain the bias toward a Th2 phenotype following treatment with Gal-1. Furthermore, increased antigen-specific Th1 and Th17 responses coupled with enhanced susceptibility to EAE was observed in Gal-1^−/− mice further supporting a role for endogenous Gal-1 in dampening Th1 and Th17 mediated pathologies [22]. Recently a Gal-1 chimeric molecule was shown to induce apoptosis of Th1 and Th17 cells obtained from the synovial fluid of RA patients further highlighting the therapeutic potential of this protein (34).

The importance of Tregs in mediating peripheral tolerance is well established and both mouse and human Tregs can produce Gal-1, furthermore the immunosuppressive capabilities of Tregs can be suppressed through neutralization of Gal-1 (32). Pertinent to this study, a role for endogenous Gal-1 was also identified as Tregs isolated from Gal-1^−/− mice exhibited a reduced suppressive capacity ex vivo. More recently, Ilarregui et al (31) highlighted another mechanism for the immunosuppressive potential of Gal-1; Gal-1 drove the generation of tolerogenic dendritic cells able to induce IL-10 producing regulatory T cells. Importantly these Gal-1-induced tolerogenic DCs could suppress inflammation in a model of EAE, an effect associated with promotion of antigen-specific T cell tolerance and blunting of Th1 and Th17 responses. As well as driving the generation of tolerogenic antigen presenting cells, a stable modified form of Gal-1 has been found to directly induce IL-10 synthesis at low micromolar concentrations in naïve T cells as well as committed Th1 and Th17 cells in a CD45-dependent manner (33). These Gal-1 treated T cells demonstrated immunosuppressive capacity both in vitro and in vivo, an effect that was abrogated when IL-10 was either neutralised or knocked out. IL-10 production has also been shown to be reduced in draining lymph nodes from Gal-1^−/− mice compared to their WT counterparts in a model of herpes induced stromal keratitis (42). It is therefore likely that absence of endogenous Gal-1 results in an enhanced immune response as demonstrated here by the enhanced disease severity in Gal-1^−/− mice combined with increased T cell proliferation and production of pro-inflammatory cytokines such as IL-17 and Il-22.

In conclusion, these data support a suppressive function for Gal-1 in autoimmune pathologies and has identified endogenous Gal-1 as a novel determinant for the arthritic response developed to CII, with important immunoregulatory properties that could be exploited for the development of novel therapies.